ИСТИНА

Background. To date the development of systems for monitoring of personal emotional arousal, EA (sleep/wake states, emotions, stress responses), are focused on remote (non-contact) methods for registration of physiological parameters. We present the data concerning the distant identification of human stress by cluster analysis terahertz (THz) radiation from different face regions, where THz signals correlated with heart rate (HR) in the situations of informational and physical stress. Methods. 38 volunteers, aged from 19 up to 57 years, participated in the experiments. Stress states were provoked by physical (threat of electric shock) or mental (solving of a hard task) stressors. For the control of EA were registered using electrocardiogram, photoplethysmogram, and galvanic skin response. The registration of the face images in the THz range by THz detector IRVT0831 was synchronous to the registration of psychophysiological indicators. Results and Discussion. The original algorithm of cluster analysis the THz radiations was developed. In stress situations there it was revealed that statistically significant (p < 0,05) resistant in time correlations between THz images, from one hand, and HR, from another one, are occurred in the areas of forehead and nasolabial triangle. It is shown that cluster analysis of distant registered THz images enables to divide the subjects into two groups by the type of reactions of the circulatory system in stressful conditions. In subjects from the first group stress causes a rush of the blood. In the subjects from the second group it provokes a spasm of the blood vessels and, as a consequence, a decrease in the intensity of blood circulation. It is also shown that the intensity of the THz signal around nose wings is reduced while person breathes. This phenomenon may be used to develop algorithms for non-contact registration of the respiratory rhythm. The authors are grateful to Dr I.A. Ozheredov and Dr A.N. Varaksin for effective assistance in conducting experiments and data analysis. The work was supported by RFBR, project № 17-29-02487.